CN116104470A - Geosteering formation identification and prediction method, system, equipment and storage medium - Google Patents
Geosteering formation identification and prediction method, system, equipment and storage mediumDownload PDFInfo
- Publication number
- CN116104470A CN116104470ACN202211413553.0ACN202211413553ACN116104470ACN 116104470 ACN116104470 ACN 116104470ACN 202211413553 ACN202211413553 ACN 202211413553ACN 116104470 ACN116104470 ACN 116104470A
- Authority
- CN
- China
- Prior art keywords
- data
- prediction
- formation
- logging
- well
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21B—EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- E21B47/00—Survey of boreholes or wells
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21B—EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- E21B49/00—Testing the nature of borehole walls; Formation testing; Methods or apparatus for obtaining samples of soil or well fluids, specially adapted to earth drilling or wells
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Forestry; Mining
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/30—Assessment of water resources
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mining & Mineral Resources (AREA)
- Geology (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Environmental & Geological Engineering (AREA)
- Fluid Mechanics (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geochemistry & Mineralogy (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Tourism & Hospitality (AREA)
- Animal Husbandry (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Marine Sciences & Fisheries (AREA)
- General Business, Economics & Management (AREA)
- Agronomy & Crop Science (AREA)
- Geophysics (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
Translated fromChinese
Description
Translated fromChinese技术领域technical field
本发明涉及石油地质勘探钻井参数识别技术领域，具体来讲，涉及一种地质导向地层识别预测方法、一种地质导向地层识别预测系统、一种实现地质导向地层识别预测方法的计算机设备和计算机存储介质。The present invention relates to the technical field of identification of drilling parameters for petroleum geological exploration, in particular to a method for identifying and predicting geosteering strata, a system for identifying and predicting geosteering strata, a computer device and computer storage for realizing the method for identifying and predicting geosteering strata medium.
背景技术Background technique
近年来，人们尝试将人工智能算法逐渐应用于油气勘探中的井间地层识别。井间地层识别是油藏描述与储层表征的关键步骤之一，进行地层识别能够了解地层层序、岩性以及地层厚度等井间结构的变化，识别结果对油藏内储集体的空间分布做出了判断，影响油气藏的开发工作。In recent years, people have attempted to gradually apply artificial intelligence algorithms to interwell formation identification in oil and gas exploration. Stratigraphic identification between wells is one of the key steps in reservoir description and reservoir characterization. Stratigraphic identification can understand the changes in interwell structures such as stratigraphic sequence, lithology, and stratum thickness. The identification results have great influence on the spatial distribution of reservoirs A judgment has been made that affects the development of oil and gas reservoirs.
地层识别对于储层油气含量以及油藏描述等方面具有重大意义。层序地层学最先是由Vail等人提出的，形成了层序地层学概念体系的雏形。现今，国内外学者对地层识别方法不断进行研究和探讨，涌现出大量的研究成果。地层识别方法主要分为地科专业识别方法、传统机器学习方法和基于深度学习的识别方法。地科专业的识别方法依赖于地球科学相关知识，通过观察、实验室分析以及人工划定等步骤对地层进行识别，如测井曲线拐点法。传统的机器学习方法包括小波变换法、BP神经网络等。目前较为流行的基于深度学习的方法能够对测井曲线数据进行自动化特征提取，应用最为广泛的是卷积神经网络方法，但这些训练方法大多的预测基础在于每一时刻使用了参数的单值，并未考虑参数变化的趋势性。Stratigraphic identification is of great significance to reservoir oil and gas content and reservoir description. Sequence stratigraphy was first proposed by Vail et al., forming the prototype of the sequence stratigraphy concept system. Nowadays, domestic and foreign scholars continue to study and discuss the method of stratum identification, and a large number of research results have emerged. Stratum identification methods are mainly divided into geoscience professional identification methods, traditional machine learning methods, and deep learning-based identification methods. The identification method of geoscience major relies on the relevant knowledge of earth science, and identifies the stratum through observation, laboratory analysis, and manual delineation, such as the inflection point method of logging curve. Traditional machine learning methods include wavelet transform method, BP neural network and so on. At present, the more popular deep learning-based methods can automatically extract features from well logging curve data. The most widely used method is the convolutional neural network method. However, most of these training methods are based on the use of single values of parameters at each moment. The trend of parameter changes is not taken into account.
例如，于2021年02月19日公开的发明名称为一种基于非开挖随钻参数机器学习的地层识别方法、公开号为CN112381938A的专利文献记载了一种基于非开挖随钻参数机器学习的地层识别方法，识别对象为未知地层，该方法的工作流程为读取大量已知地层的随钻参数，提取已知地层随钻参数的统计性特征，将特征作为属性，使用随机森林算法建立模型，最后将随钻参数导入模型识别未知地层；将未知地层的随钻参数，输入模型进行识别，得出识别结果，即未知地层的类别。于2021年07月02日公开的发明名称为一种基于非开挖泥浆流变参数弱监督机器学习的地层识别方法、公开号为CN113062734A的专利文献记载了基于非开挖泥浆流变参数弱监督机器学习的地层识别方法，识别对象为未知地层，利用弱监督机器学习与已知地层泥浆流变参数建立地层识别模型，再通过模型去识别未知地层；工作流程为读取大量已知地层的泥浆流变参数，提取已知地层泥浆流变参数的统计性特征，将特征作为属性，再使用KNN-SVM算法建立模型，最后将泥浆流变参数导入模型识别未知地层；将未知地层的泥浆流变，输入模型进行识别，得出识别结果，即未知地层的类别。这些方法的核心都在于基于地表钻机和泥浆泵检测的钻速、扭矩、转速等地表钻进参数使用随机森林、支持向量机等算法进行地层识别，预测基础在于每一时刻使用了参数的单值，并未考虑参数变化的趋势性。For example, the title of the invention published on February 19, 2021 is a stratum identification method based on machine learning of trenchless while drilling parameters. The stratum identification method, the identification object is an unknown stratum, the working process of this method is to read a large number of MWD parameters of known strata, extract the statistical characteristics of the known stratum while drilling parameters, use the characteristics as attributes, and use the random forest algorithm to establish Finally, the while-drilling parameters are imported into the model to identify unknown formations; the while-drilling parameters of unknown formations are input into the model for identification, and the identification result is obtained, that is, the category of unknown formations. The title of the invention published on July 2, 2021 is a stratum identification method based on weakly supervised machine learning of non-excavation mud rheological parameters. The stratum identification method of machine learning, the identification object is an unknown stratum, and the stratum identification model is established by using weakly supervised machine learning and the rheological parameters of the known stratum mud, and then the model is used to identify the unknown stratum; the workflow is to read a large number of mud from known strata Rheological parameters, extract the statistical characteristics of known formation mud rheological parameters, use the characteristics as attributes, and then use the KNN-SVM algorithm to build a model, and finally import the mud rheological parameters into the model to identify unknown formations; the mud rheological parameters of unknown formations , enter the model for identification, and obtain the identification result, that is, the category of the unknown formation. The core of these methods is based on surface drilling parameters detected by surface drilling rigs and mud pumps, such as drilling speed, torque, and rotational speed, using algorithms such as random forests and support vector machines to identify formations. The basis of prediction is to use a single value of parameters at each moment , without considering the trend of parameter changes.
因此，有必要形成一种考虑了参数变化的趋势性的地层识别预测方法。Therefore, it is necessary to form a stratum identification and prediction method that considers the trend of parameter changes.
发明内容Contents of the invention
本发明的目的在于解决现有技术存在的上述不足中的至少一项。例如，本发明的目的之一在于提供一种以更少的参数需求(最少仅使用一种参数)实现油气钻井过程中对地层准确识别预测的方法。The purpose of the present invention is to solve at least one of the above-mentioned deficiencies in the prior art. For example, one of the objectives of the present invention is to provide a method for accurately identifying and predicting formations during oil and gas drilling with fewer parameter requirements (only one parameter is used at least).
为了实现上述目的，本发明一方面提供了一种地质导向地层识别预测方法，所述识别预测方法包括以下步骤：S1、采集目标区域中至少一个参考井的井眼轨迹数据、测录井数据和地层数据，所述井眼轨迹数据包括井深MD、垂深TVD、大地坐标X、大地坐标Y、方位角AZIM和井斜角INCL，所述测录井数据包括井深MD和随钻伽马、成分录井和电阻率测井，所述地层数据包括地层类型和地层深度；S2、将参考井的井眼轨迹数据和测录井数据进行数据融合，所述数据融合包括数据裁剪、数据插补和数据标准化；S3、基于融合后参考井的测录井数据，利用机器学习算法建立目标区域的地层预测模型；S4、将当前预测深度之前的N组地层深度对应的预测井测录井数据和地层分类预测结果确定为滑动窗口，以所述滑动窗口作为输入参数，通过所述地层预测模型获得当前预测深度下的地层分类预测结果。In order to achieve the above object, the present invention provides a method for identifying and predicting geosteering formations. The method for identifying and predicting comprises the following steps: S1, collecting borehole trajectory data, logging data and logging data of at least one reference well in the target area. Formation data, the wellbore trajectory data includes well depth MD, vertical depth TVD, geodetic coordinate X, geodetic coordinate Y, azimuth AZIM and well inclination INCL, and the logging data includes well depth MD and gamma while drilling, composition Mud logging and resistivity logging, the formation data includes formation type and formation depth; S2, data fusion of the wellbore trajectory data of the reference well and the logging and logging data, the data fusion includes data clipping, data interpolation and Data standardization; S3. Based on the logging and mud logging data of the fused reference wells, use machine learning algorithms to establish a formation prediction model in the target area; S4. Combine the predicted well logging and mud logging data and the formation depths corresponding to the N groups of formation depths before the current predicted depth The classification prediction result is determined as a sliding window, and the sliding window is used as an input parameter to obtain the stratum classification prediction result at the current predicted depth through the stratum prediction model.
在本发明的地质导向地层识别预测方法的一个示例性实施例中，步骤S2中，在数据融合之前，可针对参考井的井眼轨迹数据和测录井数据进行数据处理，以去除异常值，所述数据处理包括离群值删除和数据平滑。In an exemplary embodiment of the geosteering formation identification and prediction method of the present invention, in step S2, before data fusion, data processing may be performed on the wellbore trajectory data and logging data of the reference well to remove abnormal values, The data processing includes outlier removal and data smoothing.
在本发明的地质导向地层识别预测方法的一个示例性实施例中，所述机器学习算法可为神经网络、支持向量机和随机森林中的一种。In an exemplary embodiment of the method for identifying and predicting geosteering formations of the present invention, the machine learning algorithm may be one of neural network, support vector machine and random forest.
在本发明的地质导向地层识别预测方法的一个示例性实施例中，步骤S2还可包括：基于融合后参考井的井眼轨迹数据，利用机器学习算法建立目标区域的测录井预测模型。In an exemplary embodiment of the method for identifying and predicting geosteering formations of the present invention, step S2 may further include: based on the fused wellbore trajectory data of the reference well, using a machine learning algorithm to establish a logging prediction model for the target area.
在本发明的地质导向地层识别预测方法的一个示例性实施例中，所述预测井测录井数据可以是将预测井的井眼轨迹数据作为输入参数，通过所述测录井预测模型获得的测录井预测结果。In an exemplary embodiment of the method for identifying and predicting geosteering formations of the present invention, the predicted well logging and mud logging data may be obtained through the logging and mud logging prediction model using the wellbore trajectory data of the predicted well as an input parameter Logging prediction results.
在本发明的地质导向地层识别预测方法的一个示例性实施例中，步骤S4中，所述预测井测录井数据可以是预测井在实际钻井过程中采集的测录井参数曲线。In an exemplary embodiment of the method for identifying and predicting geosteering formations of the present invention, in step S4, the predicted well logging data may be logging parameter curves collected during actual drilling of the predicted well.
本发明另一方面提供了一种地质导向地层识别预测系统，所述识别预测系统包括数据采集单元、数据融合单元、第一数据建模单元、第二数据建模单元、第一数据获取单元、第二数据获取单元和地层预测单元，其中，数据采集单元被配置为采集目标区域中至少一个参考井的井眼轨迹数据、测录井数据和地层数据；数据融合单元与数据采集单元连接，被配置为将参考井的井眼轨迹数据和测录井数据进行数据融合；第一数据建模单元与数据融合单元连接，被配置为基于融合后参考井的测录井数据，利用机器学习算法建立目标区域的地层预测模型；第二数据建模单元与数据融合单元连接，被配置为基于融合后参考井的井眼轨迹数据，利用机器学习算法建立目标区域的测录井预测模型；第一数据获取单元被配置为获取预测井的井眼轨迹数据；第二数据获取单元与第二数据建模单元连接，被配置为以预测井的井眼轨迹数据作为输入参数，通过所述测录井预测模型获得的测录井预测结果；地层预测单元分别与第一数据获取单元、第二数据获取单元连接，被配置为将当前预测深度之前的若干组地层深度对应的预测井的测录井预测结果和地层分类预测结果确定为滑动窗口，以所述滑动窗口作为输入参数，通过所述地层预测模型获得当前预测深度下的地层分类预测结果。Another aspect of the present invention provides a geosteering formation identification and prediction system, the identification and prediction system includes a data acquisition unit, a data fusion unit, a first data modeling unit, a second data modeling unit, a first data acquisition unit, The second data acquisition unit and formation prediction unit, wherein, the data acquisition unit is configured to acquire the borehole trajectory data, logging data and formation data of at least one reference well in the target area; the data fusion unit is connected with the data acquisition unit, and is It is configured to perform data fusion of wellbore trajectory data and mud logging data of reference wells; the first data modeling unit is connected to the data fusion unit, and is configured to establish a machine learning algorithm based on the logging and mud logging data of reference wells after fusion. The formation prediction model of the target area; the second data modeling unit is connected with the data fusion unit, and is configured to use machine learning algorithms to establish a logging prediction model of the target area based on the wellbore trajectory data of the fused reference well; the first data The acquisition unit is configured to acquire the wellbore trajectory data of the predicted well; the second data acquisition unit is connected to the second data modeling unit and is configured to use the wellbore trajectory data of the predicted well as an input parameter, and predict The logging prediction results obtained by the model; the formation prediction unit is respectively connected with the first data acquisition unit and the second data acquisition unit, and is configured to convert the logging prediction results of the prediction wells corresponding to several groups of formation depths before the current prediction depth The stratum classification prediction result is determined as a sliding window, and the stratum classification prediction result at the current predicted depth is obtained through the stratum prediction model by using the sliding window as an input parameter.
在本发明的地质导向地层识别预测系统的一个示例性实施例中，所述识别预测系统还可包括与所述地层预测单元连接的第三数据获取单元，所述第三数据获取单元被配置为获取预测井在实际钻井过程中采集的测录井参数曲线。In an exemplary embodiment of the geosteering formation identification and prediction system of the present invention, the identification and prediction system may further include a third data acquisition unit connected to the formation prediction unit, and the third data acquisition unit is configured to Obtain the logging parameter curves collected during the actual drilling of the predicted well.
本发明再一方面提供了一种计算机设备，所述计算机设备包括：处理器；和存储器，存储有计算机程序，当所述计算机程序被处理器执行时，实现如上所述的地质导向地层识别预测方法。Another aspect of the present invention provides a computer device, the computer device includes: a processor; and a memory, storing a computer program, when the computer program is executed by the processor, it realizes the above-mentioned geo-steering formation identification prediction method.
本发明再一方面提供了一种存储有计算机程序的计算机可读存储介质，当所述计算机程序在被处理器执行时实现如上所述的地质导向地层识别预测方法。Another aspect of the present invention provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the method for identifying and predicting geosteering formations as described above is implemented.
与现有技术相比，本发明的有益效果包括以下内容中的至少一项：Compared with the prior art, the beneficial effects of the present invention include at least one of the following:
(1)不同于常规算法中使用既有参数对地层进行事前预测，本发明能够在实际钻井过程中，根据实时的测录井参数曲线或者测录井预测结果，实现对实时钻井地层的准确判断，进而辅助地表钻井施工人员准确了解地下工况，避免事故发生，提高钻井施工效率；(1) Unlike conventional algorithms that use existing parameters to predict formations in advance, the present invention can accurately judge real-time drilling formations according to real-time logging parameter curves or logging prediction results during the actual drilling process , and then assist the surface drilling construction personnel to accurately understand the underground working conditions, avoid accidents, and improve drilling efficiency;
(2)本发明也能够基于同区域的参考井数据，实现新钻井的钻前地层准确预测，并根据该预测结果优选合适的钻井参数、钻井设备，并针对可能遭遇的复杂地层及其可能引发的钻井事故提前做好对应预案，进而达到提高钻井效率的目的；(2) The present invention can also realize the accurate prediction of the pre-drilling formation of the new drilling based on the reference well data in the same area, and optimize the appropriate drilling parameters and drilling equipment according to the prediction results, and aim at the complex formations that may be encountered and the possible triggers. Prepare corresponding plans in advance for drilling accidents, so as to achieve the purpose of improving drilling efficiency;
(3)本发明在预测过程中考虑了井下参数变化的趋势性，提高了预测结果的准确性。(3) The present invention considers the trend of downhole parameter changes during the prediction process, thereby improving the accuracy of the prediction results.
附图说明Description of drawings
通过下面结合附图进行的描述，本发明的上述和其他目的和/或特点将会变得更加清楚，其中：The above and other objects and/or features of the present invention will become more apparent through the following description in conjunction with the accompanying drawings, wherein:
图1示出了本发明的地质导向地层识别预测方法的一个示例性实施例的流程示意图。Fig. 1 shows a schematic flowchart of an exemplary embodiment of the geosteering formation identification and prediction method of the present invention.
图2示出了本发明的地质导向地层识别预测方法的一个示例性实施例的离群值删除结果示意图。Fig. 2 shows a schematic diagram of outlier deletion results of an exemplary embodiment of the geosteering formation identification and prediction method of the present invention.
图3示出了本发明的地质导向地层识别预测方法的一个示例性实施例的数据平滑结果示意图。Fig. 3 shows a schematic diagram of a data smoothing result of an exemplary embodiment of the geosteering formation identification and prediction method of the present invention.
图4示出了本发明的地质导向地层识别预测方法的一个示例性实施例的数据插补结果示意图。Fig. 4 shows a schematic diagram of data interpolation results of an exemplary embodiment of the geosteering formation identification and prediction method of the present invention.
图5示出了本发明的地质导向地层识别预测方法的一个示例性实施例的数据标准化结果示意图。Fig. 5 shows a schematic diagram of data normalization results of an exemplary embodiment of the geosteering formation identification and prediction method of the present invention.
图6示出了本发明的地质导向地层识别预测方法的一个示例性实施例的测录井预测模型原理示意图。Fig. 6 shows a schematic diagram of the principle of a logging prediction model of an exemplary embodiment of the geosteering formation identification and prediction method of the present invention.
图7示出了本发明的地质导向地层识别预测方法的一个示例性实施例的基于滑动窗口的地层预测模型原理示意图。Fig. 7 shows a schematic diagram of a sliding window-based formation prediction model of an exemplary embodiment of the geosteering formation identification and prediction method of the present invention.
图8A示出了本发明的地质导向地层识别预测方法的一个示例性实施例的地层测试结果示意图；图8B示出了本发明的地质导向地层识别预测方法的一个示例性实施例的。Fig. 8A shows a schematic diagram of formation test results of an exemplary embodiment of the geosteering formation identification and prediction method of the present invention; Fig. 8B shows an exemplary embodiment of the geosteering formation identification and prediction method of the present invention.
图9示出了本发明的地质导向地层识别预测系统的一个示例性实施例的预测系统结构示意图。Fig. 9 shows a schematic structural diagram of the prediction system of an exemplary embodiment of the geosteering formation identification and prediction system of the present invention.
图10示出了本发明的地质导向地层识别预测系统的一个示例性实施例的计算机设备的结构示意图。Fig. 10 shows a schematic structural diagram of the computer equipment of an exemplary embodiment of the geosteering formation identification and prediction system of the present invention.
附图标记说明：Explanation of reference signs:
10-地质导向地层识别预测系统，101-数据采集单元，102-数据融合单元，103-第一数据建模单元，104-第二数据建模单元，105-第一数据获取单元，106-第二数据获取单元，107-第三数据获取单元，108-地层预测单元，20-计算机设备、201-存储器、202-处理器。10-geo-steering formation identification and prediction system, 101-data acquisition unit, 102-data fusion unit, 103-first data modeling unit, 104-second data modeling unit, 105-first data acquisition unit, 106-the first Second data acquisition unit, 107-third data acquisition unit, 108-stratum prediction unit, 20-computer equipment, 201-memory, 202-processor.
具体实施方式Detailed ways
在下文中，将结合示例性实施例来详细说明本发明的地质导向地层识别预测方法、系统、设备和存储介质。Hereinafter, the geosteering formation identification and prediction method, system, device and storage medium of the present invention will be described in detail in combination with exemplary embodiments.
需要说明的是，“第一”、“第二”、“第三”等仅仅是为了方便描述和便于区分，而不能理解为指示或暗示相对重要性。It should be noted that "first", "second", "third", etc. are only for convenience of description and distinction, and should not be understood as indicating or implying relative importance.
本发明一方面提供了一种地质导向地层识别预测方法。One aspect of the present invention provides a geosteering formation identification and prediction method.
在本发明的一个示例性实施例中，一种地质导向地层识别预测方法包括以下步骤：In an exemplary embodiment of the present invention, a method for identifying and predicting geosteering formations includes the following steps:
步骤S1、采集目标区域中至少一个参考井的井眼轨迹数据、测录井数据和地层数据。Step S1, collecting borehole trajectory data, logging data and formation data of at least one reference well in the target area.
其中，井眼轨迹数据类型包括：井深MD、垂深TVD、大地坐标X、大地坐标Y、方位角AZIM和井斜角INCL。Among them, the borehole trajectory data types include: well depth MD, vertical depth TVD, geodetic coordinate X, geodetic coordinate Y, azimuth AZIM and well inclination INCL.
测录井数据类型包括：井深MD、随钻自然伽马、成分录井和电阻率测井等。例如，可选自然伽马，但参考井和预测井需要选用同样的参数。Logging data types include: well depth MD, natural gamma ray while drilling, composition logging and resistivity logging, etc. For example, natural gamma can be selected, but the same parameters need to be selected for reference wells and prediction wells.
地层数据类型包括：地层类型和参考井钻遇地层的地层深度。例如，100m-200m为A地层，200-500m为B地层……直到覆盖整个井眼深度。The formation data types include: formation type and formation depth encountered by reference well drilling. For example, 100m-200m is A formation, 200-500m is B formation...until the entire borehole depth is covered.
步骤S2、将参考井的井眼轨迹数据和测录井数据进行数据融合，数据融合包括数据裁剪、数据插补和数据标准化。Step S2, performing data fusion of the wellbore trajectory data of the reference well and the logging data, and the data fusion includes data clipping, data interpolation and data standardization.
数据融合的主要目的在于将参考井的井眼轨迹数据和测录井数据进行对齐和补全。正常生产中由于井眼轨迹数据、测录井数据和地层数据存在较大的数据量差异。例如，以3000m的钻井深度计算，井眼轨迹数据采样频率30m/条，轨迹数据量共100条；测录井数据0.125m/条，测录井数据24000条，地层数据整个井眼不超过10条(仅记录起始和终止深度)。因此，需要以测录井数据为核心标尺进行数据的插补，将井眼轨迹参考最接近的深度补全到测录井参数中，轨迹数据中为未出现的深度全部赋值为0或NaN，后可使用插值方法(如三次Hermite插值)进行补全。对于地层参考深度范围补全到测录井参数中，同一种地层在所属深度范围内使用相同的值进行赋值，并且出于对后续机器学习的需求，可将所有地层用数字代号进行取代。The main purpose of data fusion is to align and complete the wellbore trajectory data and logging data of reference wells. In normal production, there is a big difference in the amount of data due to borehole trajectory data, logging data and formation data. For example, based on a drilling depth of 3000m, the sampling frequency of wellbore trajectory data is 30m/piece, and the amount of trajectory data is 100 pieces; the logging data is 0.125m/piece, and the logging data is 24,000 pieces, and the formation data for the entire wellbore does not exceed 10 Bar (record start and stop depths only). Therefore, data interpolation needs to be performed with the logging data as the core scale, and the borehole trajectory is referenced to the closest depth to the logging parameters, and the depths that do not appear in the trajectory data are all assigned 0 or NaN. Afterwards, an interpolation method (such as cubic Hermite interpolation) can be used for completion. For the formation reference depth range to be added to the logging parameters, the same formation is assigned the same value within its depth range, and all formations can be replaced with digital codes for the needs of subsequent machine learning.
步骤S3、基于融合后参考井的测录井数据，利用机器学习算法建立目标区域的地层预测模型。例如，机器学习算法可为神经网络ANN、支持向量机SVM和随机森林RF中的一种，其中，随机森林RF的计算效果最佳。Step S3, based on the fused logging data of the reference well, a formation prediction model of the target area is established by using a machine learning algorithm. For example, the machine learning algorithm can be one of neural network ANN, support vector machine SVM and random forest RF, wherein random forest RF has the best calculation effect.
步骤S4、将当前预测深度之前的N组地层深度对应的预测井测录井数据和地层分类预测结果确定为滑动窗口，以滑动窗口作为输入参数，通过地层预测模型获得当前预测深度下的地层分类预测结果。Step S4. Determine the predicted well logging data and stratum classification prediction results corresponding to N groups of formation depths before the current prediction depth as a sliding window, and use the sliding window as an input parameter to obtain the formation classification at the current prediction depth through the formation prediction model forecast result.
其中，N的取值可为5～20，例如，N可取5、10、15、20等。比如，当前地层的预测输出是由当前地层前序10个输入共同参与计算完成，测井数据密度为0.125m/条，则预测10m的地层属性需要的输入参数为分别为10m、9.875m、9.75m、…、9m、8.875m的伽马数据。Wherein, the value of N may be 5-20, for example, N may be 5, 10, 15, 20 and so on. For example, the prediction output of the current stratum is calculated by the participation of 10 previous inputs of the current stratum, and the logging data density is 0.125m/strip, so the input parameters required to predict the stratum properties of 10m are 10m, 9.875m, and 9.75m respectively. Gamma data for m, ..., 9m, 8.875m.
预测井测录井数据可以是预测井在实际钻井过程中采集的测录井参数曲线。也就是说，在实际钻井施工过程中，也可以根据实际钻井采集的实际测录井参数曲线，直接根据本实施例中建立的地层预测模型进行实时的地层预测，从而实现新钻井的实时钻井地层预测。The predicted well logging data may be the logging parameter curves collected during the actual drilling of the predicted well. That is to say, in the actual drilling construction process, real-time formation prediction can also be performed directly according to the formation prediction model established in this embodiment according to the actual logging parameter curve collected by actual drilling, so as to realize the real-time drilling formation of new drilling. predict.
在本实施例中，步骤S2中，在数据融合之前，可针对参考井的井眼轨迹数据和测录井数据进行数据处理，以去除异常值，数据处理包括离群值删除和数据平滑。In this embodiment, in step S2, before data fusion, data processing may be performed on the wellbore trajectory data and logging data of the reference well to remove outliers, and the data processing includes outlier removal and data smoothing.
在本发明的另一个示例性实施例中，一种地质导向地层识别预测方法包括以下步骤：In another exemplary embodiment of the present invention, a method for identifying and predicting geosteering formations includes the following steps:
步骤S1'、采集目标区域中至少一个参考井的井眼轨迹数据、测录井数据和地层数据。Step S1', collecting borehole trajectory data, logging data and formation data of at least one reference well in the target area.
步骤S2、先将参考井的井眼轨迹数据和测录井数据进行数据处理，包括删除离群值和基本数据平滑；然后将参考井的井眼轨迹数据和测录井数据进行数据融合，数据融合包括数据裁剪、数据插补和数据标准化。Step S2, first process the wellbore trajectory data and logging data of the reference well, including deleting outliers and smoothing basic data; then perform data fusion on the wellbore trajectory data and logging data of the reference well, and the data Fusion includes data pruning, data imputation, and data normalization.
步骤S3'、基于融合后参考井的测录井数据，利用机器学习算法建立目标区域的地层预测模型；基于融合后参考井的井眼轨迹数据，利用机器学习算法建立目标区域的测录井预测模型。Step S3', based on the fused logging data of the reference well, use machine learning algorithm to establish a formation prediction model in the target area; based on the fused well trajectory data of the reference well, use machine learning algorithm to establish a logging prediction model in the target area Model.
步骤S4'、先将预测井的井眼轨迹数据作为输入参数，通过测录井预测模型获得的测录井预测结果；然后将当前预测深度之前的N组地层深度对应的预测井测录井数据(也就是测录井预测结果)和地层分类预测结果确定为滑动窗口，以滑动窗口作为输入参数，通过地层预测模型获得当前预测深度下的地层分类预测结果。Step S4', first use the wellbore trajectory data of the predicted well as an input parameter, and obtain the logging prediction result through the logging prediction model; then use the predicted logging data corresponding to the N groups of formation depths before the current predicted depth (that is, the prediction result of logging and mud logging) and the prediction result of stratum classification are determined as a sliding window, and the sliding window is used as an input parameter to obtain the prediction result of stratum classification at the current predicted depth through the formation prediction model.
在同区域进行新钻井设计时(如采集数据来源XX-1～XX-10井眼，新设计XX-11井眼)，根据钻井目标与经验获得设计井眼轨迹后，使用本实施例中的方法首先预测该井眼施工过程中可能采集的测录井数据，然后根据预测的测录井参数数据预测可能钻遇的地层，从而实现新钻井的钻前地层准确预测。When carrying out new drilling design in the same area (such as collecting data from XX-1 ~ XX-10 wellbore, newly designing XX-11 wellbore), after obtaining the designed wellbore trajectory according to the drilling target and experience, use the method in this embodiment The method first predicts the logging data that may be collected during the wellbore construction, and then predicts the formation that may be drilled according to the predicted logging parameter data, so as to realize the accurate prediction of the formation before drilling for the new well.
本发明另一方面提供了一种地质导向地层识别预测系统。Another aspect of the present invention provides a geosteering formation identification and prediction system.
在本发明的又一个示例性实施例中，地质导向地层识别预测系统包括数据采集单元、数据融合单元、第一数据建模单元、第二数据建模单元、第一数据获取单元、第二数据获取单元和地层预测单元。In yet another exemplary embodiment of the present invention, the geosteering formation identification and prediction system includes a data acquisition unit, a data fusion unit, a first data modeling unit, a second data modeling unit, a first data acquisition unit, a second data Get cells and stratigraphic prediction cells.
其中，数据采集单元被配置为采集目标区域中至少一个参考井的井眼轨迹数据、测录井数据和地层数据。Wherein, the data collection unit is configured to collect borehole trajectory data, logging data and formation data of at least one reference well in the target area.
数据融合单元与数据采集单元连接，被配置为将参考井的井眼轨迹数据和测录井数据进行数据融合。The data fusion unit is connected with the data acquisition unit and is configured to perform data fusion of wellbore trajectory data and logging data of reference wells.
第一数据建模单元与数据融合单元连接，被配置为基于融合后参考井的测录井数据，利用机器学习算法建立目标区域的地层预测模型。The first data modeling unit is connected to the data fusion unit and is configured to establish a stratum prediction model of the target area by using a machine learning algorithm based on the logging data of the fused reference well.
第二数据建模单元与数据融合单元连接，被配置为基于融合后参考井的井眼轨迹数据，利用机器学习算法建立目标区域的测录井预测模型。The second data modeling unit is connected to the data fusion unit and is configured to establish a logging prediction model of the target area by using a machine learning algorithm based on the fused wellbore trajectory data of the reference well.
第一数据获取单元被配置为获取预测井的井眼轨迹数据。The first data acquisition unit is configured to acquire borehole trajectory data of the predicted well.
第二数据获取单元与第二数据建模单元连接，被配置为以预测井的井眼轨迹数据作为输入参数，通过所述测录井预测模型获得的测录井预测结果。The second data acquisition unit is connected to the second data modeling unit and is configured to use the wellbore trajectory data of the predicted well as an input parameter to obtain the logging prediction result through the logging prediction model.
地层预测单元分别与第一数据获取单元和第二数据获取单元连接，被配置为将当前预测深度之前的若干组地层深度对应的预测井的测录井预测结果和地层分类预测结果确定为滑动窗口，以所述滑动窗口作为输入参数，通过所述地层预测模型获得当前预测深度下的地层分类预测结果。The formation prediction unit is respectively connected with the first data acquisition unit and the second data acquisition unit, and is configured to determine the logging prediction results and formation classification prediction results of the prediction wells corresponding to several groups of formation depths before the current prediction depth as a sliding window , using the sliding window as an input parameter to obtain the stratum classification prediction result at the current predicted depth through the stratum prediction model.
在本发明的又一个示例性实施例中，地质导向地层识别预测系统包括数据采集单元、数据融合单元、第一数据建模单元、第二数据建模单元、第一数据获取单元、第二数据获取单元、第三数据获取单元和地层预测单元。In yet another exemplary embodiment of the present invention, the geosteering formation identification and prediction system includes a data acquisition unit, a data fusion unit, a first data modeling unit, a second data modeling unit, a first data acquisition unit, a second data An acquisition unit, a third data acquisition unit and a stratum prediction unit.
其中，数据采集单元被配置为采集目标区域中至少一个参考井的井眼轨迹数据、测录井数据和地层数据。Wherein, the data collection unit is configured to collect borehole trajectory data, logging data and formation data of at least one reference well in the target area.
数据融合单元与数据采集单元连接，被配置为将参考井的井眼轨迹数据和测录井数据进行数据融合。The data fusion unit is connected with the data acquisition unit and is configured to perform data fusion of wellbore trajectory data and logging data of reference wells.
第一数据建模单元与数据融合单元连接，被配置为基于融合后参考井的测录井数据，利用机器学习算法建立目标区域的地层预测模型。The first data modeling unit is connected to the data fusion unit and is configured to establish a stratum prediction model of the target area by using a machine learning algorithm based on the logging data of the fused reference well.
第二数据建模单元与数据融合单元连接，被配置为基于融合后参考井的井眼轨迹数据，利用机器学习算法建立目标区域的测录井预测模型。The second data modeling unit is connected to the data fusion unit and is configured to establish a logging prediction model of the target area by using a machine learning algorithm based on the fused wellbore trajectory data of the reference well.
第一数据获取单元被配置为获取预测井的井眼轨迹数据。The first data acquisition unit is configured to acquire borehole trajectory data of the predicted well.
第二数据获取单元与第二数据建模单元连接，被配置为以预测井的井眼轨迹数据作为输入参数，通过所述测录井预测模型获得的测录井预测结果。The second data acquisition unit is connected to the second data modeling unit and is configured to use the wellbore trajectory data of the predicted well as an input parameter to obtain the logging prediction result through the logging prediction model.
第三数据获取单元被配置为获取预测井在实际钻井过程中采集的测录井参数曲线。The third data acquisition unit is configured to acquire the logging parameter curves collected during actual drilling of the predicted well.
地层预测单元分别与第一数据获取单元、第二数据获取单元和第三数据获取单元连接，被配置为将当前预测深度之前的若干组地层深度对应的预测井的测录井预测结果或测录井参数曲线、以及地层分类预测结果确定为滑动窗口，以所述滑动窗口作为输入参数，通过所述地层预测模型获得当前预测深度下的地层分类预测结果。The stratum prediction unit is respectively connected with the first data acquisition unit, the second data acquisition unit and the third data acquisition unit, and is configured to collect the logging prediction results or logging prediction results of the prediction wells corresponding to several groups of formation depths before the current prediction depth. The well parameter curve and the formation classification prediction result are determined as a sliding window, and the formation classification prediction result at the current predicted depth is obtained through the formation prediction model using the sliding window as an input parameter.
根据本发明的地质导向地层识别预测方法可以被编程为计算机程序并且相应的程序代码或指令可以被存储在计算机可读存储介质中，当程序代码或指令被处理器执行时使得处理器执行上述地质导向地层识别预测方法，上述处理器和存储器可以被包括在计算机设备中。The method for identifying and predicting geosteering formations according to the present invention can be programmed as a computer program and the corresponding program codes or instructions can be stored in a computer-readable storage medium. When the program codes or instructions are executed by a processor, the processor can perform the above-mentioned geological For the method of guiding formation identification and prediction, the above processor and memory may be included in computer equipment.
根据本发明又一方面的示例性实施例还提供了一种存储有计算机程序的计算机可读存储介质。该计算机可读存储介质存储有当被处理器执行时使得处理器执行根据本发明的地质导向地层识别预测方法的计算机程序。该计算机可读记录介质是可存储由计算机系统读出的数据的任意数据存储装置。计算机可读记录介质的示例包括：只读存储器、随机存取存储器、只读光盘、磁带、软盘、光数据存储装置和载波(诸如经有线或无线传输路径通过互联网的数据传输)。Exemplary embodiments according to yet another aspect of the present invention also provide a computer-readable storage medium storing a computer program. The computer-readable storage medium stores a computer program that, when executed by a processor, causes the processor to execute the method for identifying and predicting geosteering formations according to the present invention. The computer-readable recording medium is any data storage device that can store data read by a computer system. Examples of computer-readable recording media include: read-only memory, random-access memory, optical disc, magnetic tape, floppy disk, optical data storage devices, and carrier waves (such as data transmission over the Internet via wired or wireless transmission paths).
根据本发明又一方面的示例性实施例还提供了一种计算机设备。该计算机设备包括处理器和存储器。存储器用于存储计算机程序。计算机程序被处理器执行使得处理器执行根据本发明的地质导向地层识别预测方法的计算机程序。Exemplary embodiments according to yet another aspect of the present invention also provide a computer device. The computer device includes a processor and memory. Memory is used to store computer programs. The computer program is executed by the processor so that the processor executes the computer program of the geosteering formation identification and prediction method according to the present invention.
为了更好地理解本发明的上述示例性实施例，下面结合附图和具体示例对其进行进一步说明。In order to better understand the above exemplary embodiments of the present invention, it will be further described below in conjunction with the accompanying drawings and specific examples.
示例1Example 1
一种智能地质导向地层识别预测方法的整体预测流程如图1所示，整体划分为针对区域性数据采集、处理、融合和建模预测四个步骤。The overall prediction process of an intelligent geosteering stratum identification and prediction method is shown in Figure 1, which is divided into four steps for regional data collection, processing, fusion and modeling prediction.
各步骤详解如下：Each step is explained in detail as follows:
步骤一、数据采集。采集区域性参考井数据，主要包含井眼轨迹、拟使用的测录井数据(如自然伽马)、参考地层分类数据(即地层数据)。例如，在指定区域需要施工井号为XX-11的钻井，则需要事先采集到同区域井号XX-1、XX-2……XX-10(并不一定需要该区域所有钻井，但出于准确性考虑，数据越多越好)的井眼轨迹数据、测录井数据、地层数据。Step one, data collection. Collect regional reference well data, mainly including wellbore trajectory, logging data to be used (such as natural gamma ray), and reference formation classification data (ie formation data). For example, if the well number XX-11 needs to be drilled in a designated area, it is necessary to collect well numbers XX-1, XX-2...XX-10 in the same area in advance (not necessarily all the wells in this area are required, but for Considering accuracy, the more data the better), borehole trajectory data, logging data, formation data.
其中，井眼轨迹数据包括井眼斜深、垂深、井斜角、方位角、大地坐标X和Y。实钻测录井数据曲线视需要而定(例如可选自然伽马，但参考井和预测井需要选用同样的参数)。地层分类数据主要包括参考井钻遇地层的深度区间(如100m-200m为A地层，200-500m为B地层……直到覆盖整个井眼深度)。Wherein, the borehole trajectory data includes borehole inclination depth, vertical depth, inclination angle, azimuth angle, geodetic coordinates X and Y. The actual drilling and logging data curve is determined according to the needs (for example, natural gamma ray can be selected, but the same parameters need to be selected for reference wells and prediction wells). The formation classification data mainly includes the depth interval of the formation encountered by the reference well (for example, 100m-200m is the A formation, 200-500m is the B formation...until the entire borehole depth is covered).
步骤二、数据处理。对采集的参考井数据进行数据处理，包括删除离群值、基本数据平滑等操作。Step two, data processing. Perform data processing on the collected reference well data, including operations such as deleting outliers and smoothing basic data.
举例如下：离群值删除如式1所示，计算Z-score得分Z，将|Z|≥3值的初始数据视为离群值删除，经过删除前后的数据对比如图2所示。An example is as follows: Outlier deletion is shown in
式中，xi—初始值；
数据平滑：基于SG平滑滤波器(Savizky-Golay)，选用窗口大小为10，滤波多项式为2，滤波效果如图3所示。Data smoothing: Based on the SG smoothing filter (Savizky-Golay), the selected window size is 10, and the filtering polynomial is 2. The filtering effect is shown in Figure 3.
步骤三、数据融合，包括数据裁剪、数据插补和数据标准化。该步骤的主要目的在于将参考井井眼轨迹数据和测录井数据进行对齐和补全。Step 3, data fusion, including data clipping, data imputation and data standardization. The main purpose of this step is to align and complete the reference well borehole trajectory data and logging data.
①数据裁剪：如下表1所示，井斜数据与随钻测井伽马曲线数据并不匹配，需要以伽马数据为准(由于伽马数据量小)进行裁剪匹配。①Data clipping: As shown in Table 1 below, the well deviation data does not match the logging-while-drilling gamma curve data, and the gamma data needs to be used as the standard (due to the small amount of gamma data) for clipping and matching.
②数据插补：井斜数据与随钻测井伽马曲线采样密度不匹配，需要对大间隔的井斜数据进行插补，使其具有和伽马曲线同样的间隔，使用了三次Hermite插值的方法进行，插值效果如图4所示，插值前数据为74条，插值后数据为16763条。②Data interpolation: The sampling density of the well deviation data does not match the logging-while-drilling gamma curve. It is necessary to interpolate the well deviation data at large intervals to make it have the same interval as the gamma curve. The three-time Hermite interpolation method is used. method, the interpolation effect is shown in Figure 4, the data before interpolation is 74, and the data after interpolation is 16763.
③数据标准化：使用min-max算法(式2)进行对应的数据标准化，标准化后将数据整体转化为[0,1]区间。考虑到不同的井眼和潜在的不同测量设备，在标准化过程中将基于不同的井眼进行对应的标准化。数据标准化结果如图5所示。③ Data standardization: Use the min-max algorithm (Formula 2) to standardize the corresponding data, and convert the data as a whole into the [0,1] interval after standardization. Considering different boreholes and potentially different measurement equipment, the corresponding standardization will be carried out based on different boreholes during the standardization process. The results of data normalization are shown in Figure 5.
xn＝(xi-Xmin)/(Xmax-Xmin) (2)xn =(xi -Xmin )/(Xmax -Xmin ) (2)
式中，Xmin/Xmax—初始数据集的极小值/极大值。In the formula, Xmin /Xmax —the minimum value/maximum value of the initial data set.
表1井斜数据与随钻测井伽马曲线数据Table 1 Well deviation data and LWD gamma curve data
步骤四、数据预测。分别建立基于井眼轨迹参数的测录井参数预测模型和基于测录井参数的地层预测模型，进而实现在完成井眼轨迹设计后直接进行本区域的地层预测。所建模型的步骤如下：Step four, data prediction. The logging parameter prediction model based on the wellbore trajectory parameters and the formation prediction model based on the logging parameters are respectively established, so as to realize the formation prediction in this area directly after the completion of the wellbore trajectory design. The steps to build the model are as follows:
①建立本区域的测录井参数模型，基于采集的同区域参考井数据，经过数据融合后，选择合适的机器学习算法，直接建立井眼轨迹和测录井参数的预测模型，如图6所示。①Establish the logging parameter model in this area, based on the collected reference well data in the same area, after data fusion, select the appropriate machine learning algorithm, and directly establish the prediction model of wellbore trajectory and logging parameters, as shown in Fig. 6 Show.
伽马预测属于回归预测，可以选用MSE和R2作为伽马预测指标值进行评估，如式3、式4所示。需要说明的是，机器学习算法可考虑不同的机器学习模型，例如，神经网络NN、支持向量机SVM、随机森林RF等，三种模型的训练效果如表2所示。Gamma prediction belongs to regression prediction, and MSE andR2 can be selected as gamma prediction index values for evaluation, as shown in Equation 3 and Equation 4. It should be noted that the machine learning algorithm can consider different machine learning models, for example, neural network NN, support vector machine SVM, random forest RF, etc. The training effects of the three models are shown in Table 2.
表2不同算法的伽马预测对比结果Table 2 Gamma prediction comparison results of different algorithms
需要说明的是，在此步骤的输入参数为井眼轨迹参数，包括井深MD、垂深TVD、大地坐标X、大地坐标Y、方位角AZIM和井斜角INCL，输出参数为测录井参数。本示例中给出的测录井参数为随钻伽马，但并不局限与此，后期可根据现场实际钻井情况进行更改。确定输入输出参数后调用不同的模型进行训练，本示例使用了神经网络、支持向量机和随机森林，最后比较预测效果，认为随机森林的训练效果最佳。It should be noted that the input parameters in this step are wellbore trajectory parameters, including well depth MD, vertical depth TVD, geodetic coordinates X, geodetic coordinates Y, azimuth AZIM and well inclination INCL, and the output parameters are logging parameters. The logging and logging parameters given in this example are gamma while drilling, but not limited thereto, and can be changed later according to the actual drilling conditions on site. After determining the input and output parameters, call different models for training. In this example, neural network, support vector machine and random forest are used. Finally, the prediction effect is compared, and the training effect of random forest is considered to be the best.
②建立本区域的地层预测模型。基于采集的同区域参考井数据，经过数据融合后，选择合适的机器学习算法，直接建立地层预测模型，利用测录井参数预测地层。②Establish the formation prediction model in this area. Based on the collected reference well data in the same area, after data fusion, an appropriate machine learning algorithm is selected to directly establish a formation prediction model, and the formation is predicted using logging parameters.
需要说明的是，在此步骤的输入参数为测录井参数，本示例中给出的测录井参数为随钻伽马，并不局限与此，后期可根据现场实际钻井情况进行更改，输出参数为地层类型。确定输入输出参数后调用不同的模型进行训练，本示例使用了神经网络、支持向量机和随机森林，最后比较预测效果，认为随机森林的训练效果最佳。It should be noted that the input parameters in this step are logging parameters. The logging parameters given in this example are gamma while drilling, which is not limited to this. It can be changed later according to the actual drilling conditions on site. The output The parameter is the stratum type. After determining the input and output parameters, call different models for training. In this example, neural network, support vector machine and random forest are used. Finally, the prediction effect is compared, and the training effect of random forest is considered to be the best.
地层预测的核心在于将随钻测录井数据作为输入数据引入模型进行预测，与传统多输入单输出的预测模型不同，在此使用了基于滑动窗口的预测方式。以仅使用随钻伽马数据为例(实际也可使用多种参数，例如，随钻测录井的所有参数，包括但不限于如成分录井、电阻率测井等参数)，虽然输入参数只有伽马值一种，但将预测深度之前若干伽马值及其变化趋势作为输入，同时将预测地层分类结果也引入进行预测，预测算法如图7所示。图7中的符号A表示伽马数据，符号B表示地层数据，N表示滑动窗口。当前地层的预测输出是由当前地层前序10个输入共同参与计算完成。本文使用数据集测井数据密度0.125m/条，则预测10m的地层属性y(t＝10)需要的输入参数为分别为10m、9.875m、9.75m、…、9m、8.875m的伽马数据。The core of formation prediction is to import logging-while-drilling and mud logging data as input data into the model for prediction. Unlike the traditional multi-input and single-output prediction model, a prediction method based on sliding windows is used here. Take gamma data while drilling as an example (in fact, various parameters can also be used, for example, all parameters of logging while drilling, including but not limited to parameters such as composition logging, resistivity logging, etc.), although the input parameters There is only one kind of gamma value, but several gamma values and their change trends before the predicted depth are used as input, and the predicted stratigraphic classification results are also introduced for prediction. The prediction algorithm is shown in Figure 7. Symbol A in Fig. 7 represents gamma data, symbol B represents stratigraphic data, and N represents a sliding window. The prediction output of the current stratum is calculated by the joint participation of the previous 10 inputs of the current stratum. In this paper, the logging data density of the data set is 0.125m/strip, so the input parameters needed to predict the formation property y (t=10) of 10m are the gamma data of 10m, 9.875m, 9.75m, ..., 9m, and 8.875m .
换句话说，当输入参数仅为随钻伽马数据这一种参数时，将这一种参数经过N＝10的滑动窗口取值后变成10列输入，以避免输入参数过少带来的误差；当输入参数包括随钻伽马、成分录井、电阻率测井等三种参数时，将这三种参数经过N＝10的滑动窗口取值后变成3×10列的矩阵输入。In other words, when the input parameter is only a parameter of gamma data while drilling, this parameter is converted into 10 columns of input after the sliding window value of N=10, so as to avoid the error caused by too few input parameters. Error: When the input parameters include three parameters such as gamma while drilling, composition logging, and resistivity logging, these three parameters are converted into a matrix input of 3×10 columns after taking values through a sliding window of N=10.
地层预测属于分类预测，可以选用准确率进行相关判定，如式3所示，计算结果ACC代表了分类正确的样本数TP占整个数据集样本数n的比例。Stratigraphic prediction belongs to classification prediction, and the accuracy rate can be selected for related judgments. As shown in Equation 3, the calculation result ACC represents the ratio of the number of correctly classified samples TP to the number of samples n in the entire data set.
三种模型的训练效果如表3所示，三种算法的测试结果如表3所示，实际预测效果如图8A和图8B所示(为确保测试效果，从训练数据中分别提取了对应数量的不同地层参与测试，测试数据量占总数据量20％)。The training effects of the three models are shown in Table 3, the test results of the three algorithms are shown in Table 3, and the actual prediction effects are shown in Figure 8A and Figure 8B (in order to ensure the test effect, the corresponding numbers were extracted from the training data Different strata of different strata participated in the test, and the test data volume accounted for 20% of the total data volume).
表3不同算法的地层分类预测对比结果Table 3 Comparison results of stratigraphic classification and prediction of different algorithms
整体算法的实际使用，在同区域进行新钻井设计时(如采集数据来源XX-1～XX-10井眼，新设计XX-11井眼)，根据钻井目标与经验获得设计井眼轨迹后，可直接使用本方法首先预测该井眼施工过程中可能采集的测录井参数数据，后根据预测的测录井参数数据预测可能钻遇的地层，在实际钻井施工过程中，也可以根据实际钻井采集的实际测录井参数曲线，直接根据本方法建立的地层预测模型进行实时的地层预测。For the actual use of the overall algorithm, when designing a new well in the same area (such as collecting data from XX-1 ~ XX-10 wellbore and newly designing XX-11 wellbore), after obtaining the designed wellbore trajectory according to the drilling objectives and experience, This method can be directly used to first predict the logging parameter data that may be collected during the wellbore construction, and then predict the formation that may be drilled according to the predicted logging parameter data. In the actual drilling construction process, it can also be based on the actual drilling. The collected actual logging parameter curves are directly used for real-time formation prediction based on the formation prediction model established by this method.
本示例基于同区域的参考井数据，可实现新钻井的钻前地层准确预测，并根据该预测结果优选合适的钻井参数、钻井设备，并针对可能遭遇的复杂地层及其可能引发的钻井事故提前做好对应预案，进而达到提高钻井效率的目的。当然，在实际钻井过程中，也可根据实时的测录井参数曲线，实现对实时钻井地层的准确判断，进而辅助地表钻井施工人员准确了解地下工况，避免事故发生，提高钻井施工效率。This example is based on the reference well data in the same area, which can realize the accurate prediction of the pre-drilling formation of the new drilling, and optimize the appropriate drilling parameters and drilling equipment according to the prediction results, and advance the drilling accidents that may be encountered in complex formations that may be encountered. Prepare corresponding plans to achieve the goal of improving drilling efficiency. Of course, in the actual drilling process, the real-time logging parameter curve can also be used to accurately judge the real-time drilling formation, and then assist the surface drilling personnel to accurately understand the underground working conditions, avoid accidents, and improve drilling efficiency.
示例2Example 2
如图9所示，地质导向地层识别预测系统10包括数据采集单元101、数据融合单元102、第一数据建模单元103、第二数据建模单元104、第一数据获取单元105、第二数据获取单元106、第三数据获取单元107和地层预测单元108。As shown in Figure 9, the geosteering formation identification and
其中，数据采集单元101被配置为采集目标区域中至少一个参考井的井眼轨迹数据、测录井数据和地层数据。Wherein, the
数据融合单元102与数据采集单元101连接，被配置为将参考井的井眼轨迹数据和测录井数据进行数据融合。The
第一数据建模单元103与数据融合单元102连接，被配置为基于融合后参考井的测录井数据，利用机器学习算法建立目标区域的地层预测模型。The first
第二数据建模单元104与数据融合单元102连接，被配置为基于融合后参考井的井眼轨迹数据，利用机器学习算法建立目标区域的测录井预测模型。The second
第一数据获取单元105被配置为获取预测井的井眼轨迹数据。The first
第二数据获取单元106与第二数据建模单元104连接，被配置为以预测井的井眼轨迹数据作为输入参数，通过所述测录井预测模型获得的测录井预测结果。The second
第三数据获取单元107被配置为获取预测井在实际钻井过程中采集的测录井参数曲线。The third
地层预测单元108分别与第一数据获取单元105、第二数据获取单元106、第三数据获取单元107连接，被配置为将当前预测深度之前的若干组地层深度对应的预测井的测录井预测结果或测录井参数曲线、以及地层分类预测结果确定为滑动窗口，以所述滑动窗口作为输入参数，通过所述地层预测模型获得当前预测深度下的地层分类预测结果。The
示例3Example 3
如图10所示，一种计算机设备20包括存储器201和处理器202，存储器201中存储有计算机程序，当所述计算机程序被处理器执行时，实现如上所述的地质导向地层识别预测方法。As shown in FIG. 10 , a
综上所述，本发明的有益效果包括以下内容中的至少一项：In summary, the beneficial effects of the present invention include at least one of the following:
(1)不同于常规算法中使用既有参数对地层进行事前预测，本发明能够在实际钻井过程中，根据实时的测录井参数曲线或者测录井预测结果，实现对实时钻井地层的准确判断，进而辅助地表钻井施工人员准确了解地下工况，避免事故发生，提高钻井施工效率；(1) Unlike conventional algorithms that use existing parameters to predict formations in advance, the present invention can accurately judge real-time drilling formations according to real-time logging parameter curves or logging prediction results during the actual drilling process , and then assist the surface drilling construction personnel to accurately understand the underground working conditions, avoid accidents, and improve drilling efficiency;
(2)本发明也能够基于同区域的参考井数据，实现新钻井的钻前地层准确预测，并根据该预测结果优选合适的钻井参数、钻井设备，并针对可能遭遇的复杂地层及其可能引发的钻井事故提前做好对应预案，进而达到提高钻井效率的目的；(2) The present invention can also realize the accurate prediction of the pre-drilling formation of the new drilling based on the reference well data in the same area, and optimize the appropriate drilling parameters and drilling equipment according to the prediction results, and aim at the complex formations that may be encountered and the possible triggers. Prepare corresponding plans in advance for drilling accidents, so as to achieve the purpose of improving drilling efficiency;
(3)本发明在预测过程中考虑了井下参数变化的趋势性，提高了预测结果的准确性。(3) The present invention considers the trend of downhole parameter changes during the prediction process, thereby improving the accuracy of the prediction results.
尽管上面已经结合示例性实施例及附图描述了本发明，但是本领域普通技术人员应该清楚，在不脱离权利要求的精神和范围的情况下，可以对上述实施例进行各种修改。Although the invention has been described above with reference to the exemplary embodiments and accompanying drawings, it will be apparent to those skilled in the art that various modifications may be made to the above-described embodiments without departing from the spirit and scope of the claims.
Claims (10)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211413553.0ACN116104470A (en) | 2022-11-11 | 2022-11-11 | Geosteering formation identification and prediction method, system, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211413553.0ACN116104470A (en) | 2022-11-11 | 2022-11-11 | Geosteering formation identification and prediction method, system, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116104470Atrue CN116104470A (en) | 2023-05-12 |
Family
ID=86262719
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211413553.0APendingCN116104470A (en) | 2022-11-11 | 2022-11-11 | Geosteering formation identification and prediction method, system, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116104470A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118309408A (en)* | 2024-04-01 | 2024-07-09 | 北京蓝海智信能源技术有限公司 | Method, system, computer equipment and storage medium for determining drilling engineering parameters |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150088424A1 (en)* | 2013-09-20 | 2015-03-26 | Schlumberger Technology Corporation | Identifying geological formation depth structure using well log data |
| CN110020446A (en)* | 2017-08-30 | 2019-07-16 | 中国石油化工股份有限公司 | A kind of geosteering method for early warning |
| US20200109618A1 (en)* | 2018-10-05 | 2020-04-09 | Ubiterra Corporation | Systems and methods for geosteering during well drilling |
| RU2720115C1 (en)* | 2018-01-24 | 2020-04-24 | Общество с ограниченной ответственностью "Геонавигационные технологии" | Method of automated geological survey of wells and system for its implementation |
| CN113159099A (en)* | 2021-02-09 | 2021-07-23 | 北京工业大学 | Atmospheric pollutant concentration prediction method based on multi-task learning |
| US20210326721A1 (en)* | 2020-04-17 | 2021-10-21 | Quantico Energy Solutions Llc | Earth modeling methods using machine learning |
| CN114086887A (en)* | 2021-11-19 | 2022-02-25 | 东北石油大学 | Downhole planning method for track of borehole to be drilled based on artificial intelligence |
| CN114742964A (en)* | 2022-05-07 | 2022-07-12 | 中国石油大学(北京) | Three-dimensional space expansion modeling method, device, equipment and storage medium for well logging data |
- 2022
- 2022-11-11CNCN202211413553.0Apatent/CN116104470A/enactivePending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150088424A1 (en)* | 2013-09-20 | 2015-03-26 | Schlumberger Technology Corporation | Identifying geological formation depth structure using well log data |
| CN110020446A (en)* | 2017-08-30 | 2019-07-16 | 中国石油化工股份有限公司 | A kind of geosteering method for early warning |
| RU2720115C1 (en)* | 2018-01-24 | 2020-04-24 | Общество с ограниченной ответственностью "Геонавигационные технологии" | Method of automated geological survey of wells and system for its implementation |
| US20200109618A1 (en)* | 2018-10-05 | 2020-04-09 | Ubiterra Corporation | Systems and methods for geosteering during well drilling |
| US20210326721A1 (en)* | 2020-04-17 | 2021-10-21 | Quantico Energy Solutions Llc | Earth modeling methods using machine learning |
| CN113159099A (en)* | 2021-02-09 | 2021-07-23 | 北京工业大学 | Atmospheric pollutant concentration prediction method based on multi-task learning |
| CN114086887A (en)* | 2021-11-19 | 2022-02-25 | 东北石油大学 | Downhole planning method for track of borehole to be drilled based on artificial intelligence |
| CN114742964A (en)* | 2022-05-07 | 2022-07-12 | 中国石油大学(北京) | Three-dimensional space expansion modeling method, device, equipment and storage medium for well logging data |
Non-Patent Citations (1)
| Title |
|---|
| 李建民;李黔;梁海波;张继军;乐守群;: "煤层气水平井的煤层实时识别技术", 天然气工业, no. 10, 25 October 2010 (2010-10-25), pages 60 - 63* |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118309408A (en)* | 2024-04-01 | 2024-07-09 | 北京蓝海智信能源技术有限公司 | Method, system, computer equipment and storage medium for determining drilling engineering parameters |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104948176B (en) | A kind of method based on infiltration Magnification identification carbonate reservoir crack | |
| US9176255B2 (en) | Permeability prediction systems and methods using quadratic discriminant analysis | |
| CN110397402B (en) | Drilling method and device | |
| CN109085663A (en) | A kind of tight sandstone reservoir stratification seam recognition methods | |
| CN107133670A (en) | A kind of Complex lithologic identification method and system based on decision tree data mining algorithm | |
| CN107861917B (en) | Method of the multivariate data combined calculation wellbore to stratigraphic boundary distance in horizontal well | |
| CN116305850B (en) | A method for predicting formation thermal conductivity based on random forest model | |
| CN102243672B (en) | Gushing operation condition soft sensing modeling method based on hybrid multiple models in shield tunneling process | |
| CN116451013B (en) | A Method for In-Situ Drillability Level Prediction of Deep Formation Rocks | |
| CN107895092B (en) | Inter-well communication quantitative evaluation method based on complex nonlinear injection-production modeling | |
| CN114943015A (en) | Oil reservoir analogy method and device based on similarity calculation model | |
| CN104153768A (en) | Granite reservoir stratum reservoir performance evaluation method | |
| Pang et al. | Machine learning for carbonate formation drilling: Mud loss prediction using seismic attributes and mud loss records | |
| CN116104470A (en) | Geosteering formation identification and prediction method, system, equipment and storage medium | |
| CN114818493B (en) | A method for quantitative evaluation of tunnel rock mass integrity | |
| CN110646847A (en) | Method for identifying low-order fault breakpoint position based on well logging data of close well pattern area | |
| Taipova et al. | Verifying reserves opportunities with multi-well pressure pulse-code testing | |
| CN116335648A (en) | Quantitative evaluation method of sweet spot quality in conglomerate horizontal wells based on logging-while-drilling data | |
| Suarez et al. | Fracturing-to-Production Simulation Approach for Completion Optimization in the Vaca Muerta Shale | |
| CN117846588A (en) | Method for determining formation lithology of fourth century by using susceptibility logging method | |
| WO2024087800A1 (en) | Conventional well logging data-based method for identifying high-frequency cyclic carbonate rock | |
| CN113431557B (en) | Underground borehole track tracking method based on artificial intelligence | |
| CN116304921A (en) | An Intelligent Interpretation Method for Reservoir Diagenetic Facies Logging | |
| CN109598049A (en) | Method for drilling rock fracture development degree and regional rock fracture development rule | |
| CN116029550A (en) | A risk assessment method for horizontal well geosteering with well-seismic integration |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |